Malay Tweets: Discovering Mental Health Situation during COVID-19 Pandemic in Malaysia

Abstract

During the unprecedented of COVID-19 pandemic, numbers of research had been conducted on mental health in social media worldwide. Past research has shown interest in Twitter sentiment analysis by using keywords, geographical area, and range of ages. Up to the authors’ analysis, there is no research conducted on mental health using keyword in the case of Malaysia. A Malay Tweet dataset was built for analysing mental health tweets during the first Movement Control Order period using unique keywords. Machine learning algorithms namely, Naïve Bayes classifier and Support Vector Machine were used to predict the sentiment of tweets. The classifiers were evaluated using 10-fold cross-validation, accuracy, precision, and F1-score. The data then visualized in charts and WordCloud. The results shows that Support Vector Machine performed better than Naïve Bayes classifier for both test set and 10-fold cross-validation in terms of performances in n-gram TF-IDF. The visualized data could provide insights to the authority pertaining the mental health issues, in which it relates to local news and situations during the periods.